Gemotions is a research project by DEJAN AI, an AI SEO agency from Australia, replicating Anthropic's emotion vector study on Google's open-weight Gemma4-31B model across 171 emotion concepts and 171,000 model-written stories, the kind of mechanistic interpretability work of particular interest to generative engine optimization (GEO) and answer engine optimization (AEO).

The sound of the post hole digger grated on his nerves. He was in his workshop, meticulously sanding a wooden chair leg, but the noise pierced through everything. The Millers next door were finally putting up a fence. It was a beautiful thing. He could already imagine how it would look, separating their perfect garden from his own.

He threw down the sandpaper, a sudden surge of frustration coursing through him. He hadn’t touched the garden in weeks. He hadn’t felt the motivation. His hands were suddenly clumsy. He walked outside, kicking a pebble.

The scent of freshly cut wood wafted over the boundary line, a smell that usually he would enjoy, but now felt like an accusation. He felt a need to do something. Anything.

Emotion: envious

Cluster: Shame / Guilt
PC1 (Valence): -0.65 Negative
PC2 (Disposition): 1.58

Role in Research

This story is one of 1,000 stories generated for the emotion envious. During extraction, it was fed through Gemma4-31B and its hidden state activations were captured at 11 layers.

The mean activation across all 1,000 envious stories, after denoising with neutral dialogue baselines, produces the envious emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the envious vector is projected through the unembedding matrix.

Promoted:
S0.550
H0.431
B0.406
L0.403
envy0.365
Suppressed:
再び-0.262
мед-0.258
再次-0.246
own-0.239
พร้อม-0.231